[HN Gopher] Design For Reliability
       ___________________________________________________________________
        
       Design For Reliability
        
       Author : bryanrasmussen
       Score  : 66 points
       Date   : 2021-03-14 06:07 UTC (1 days ago)
        
 (HTM) web link (semiengineering.com)
 (TXT) w3m dump (semiengineering.com)
        
       | ChrisMarshallNY wrote:
       | I'm glad to see this. Not sure if anyone's listening, though.
       | Reliability and robustness seem to have fallen by the wayside, in
       | a rush for money.
       | 
       | One of my pet peeves, is Bluetooth devices that crap out, after a
       | year.
       | 
       | I've gone through quite a few expensive headsets, while my cheap
       | exercise headsets have lasted for many years.
        
         | blacktriangle wrote:
         | There's an ebb and flow to these things. If new technology is
         | making my headset obsolete every year, designing a 10-year
         | headset makes no sense. Once headset technology settles down
         | and people realize they keep buying the same thing over and
         | over, then the customer starts to value reliability more.
         | Unfortunately customer behavior updating to the reality that
         | underlying technology of a product has stopped evolving lags by
         | a few years, but it happens.
        
           | ChrisMarshallNY wrote:
           | Like flatbed scanners.
           | 
           | Most are still USB 2, and the resolution is still 600 X 600
           | (like it was ten years ago). I have an old HP that I got a
           | decade ago, that basically still has the same specs as their
           | current flagship.
        
           | tablespoon wrote:
           | > There's an ebb and flow to these things. If new technology
           | is making my headset obsolete every year, designing a 10-year
           | headset makes no sense. Once headset technology settles down
           | and people realize they keep buying the same thing over and
           | over, then the customer starts to value reliability more.
           | Unfortunately customer behavior updating to the reality that
           | underlying technology of a product has stopped evolving lags
           | by a few years, but it happens.
           | 
           | However, business _really like_ the income that comes from
           | regular tech refreshes. Once technology stops making their
           | products obsolete after N years, they 'll often start
           | designing them to reduce reliability (e.g. going for cheap
           | parts that will fail faster) or incorporating planned
           | obsolescence features.
           | 
           | IIRC, this is what's happened to consumer printers and many
           | types of home appliances.
        
         | mumblemumble wrote:
         | I don't think this is deliberate, but it seems that one of the
         | big things that has changed as a result of the shortening of
         | consumer product development cycles is that it sort of
         | eliminates any feedback loop that might incentivize
         | manufacturers to engineer for reliability. Why build a device
         | that's designed to last for 10 years, when consumers are going
         | to buy themselves a new one every year or two, regardless of
         | whether the old one is still functional and meeting their
         | needs?
        
       | taxicabjesus wrote:
       | Recently my father had wifi problems at his house. I thought the
       | smart thermostat had a firmware update that was flawed, resulting
       | in it not staying on the network for more than a day. After a few
       | months of this all the other devices started losing their wifi
       | connections too. I replaced the Ubiquiti access point with a
       | router flashed with ddwrt [0], and everything now works fine.
       | Maybe a neighbor's wifi network started overlapping on the
       | channel, but I think it most likely the access point just died.
       | It was mounted where it got sun every morning, maybe the heat
       | cycles accelerated the AP's aging?
       | 
       | I bought a parts laptop off ebay to fix the broken hinge on the
       | HP laptop I bought in the summer of 2012. After getting my laptop
       | working, I figured the parts laptop would probably work if I
       | bought a replacement "DC IN Power Jack". Glued the broken hinge
       | back together, bought the jack off ebay, and _poof_ "now I have
       | two laptops". I put the second one on a network cable and used it
       | as a media server.
       | 
       | In recent months I've basically stopped using this laptop because
       | the wifi is so unreliable. The donor laptop's wifi seems to be
       | unreliable too. My 2012-2014 era Toshiba laptop usually works,
       | but every so often its wifi craps out on me too.
       | 
       | What's your experience with aging WiFi chips? The problem with
       | trying to fix things is if you don't have the experience, tools,
       | aptitude or a collection parts, fixing something is often a
       | mission-impossible. It's usually easier to throw out the "broken
       | laptop", when all the device actually needs is a replaceable
       | part. (I'm going to ifixit to look at the repairability score
       | before buying another laptop.)
       | 
       | [0] https://openwrt.org/meta/infobox/broadcom_wifi - bought the
       | router at a thrift store for $2, on account of it being listed as
       | supported by ddwrt. ddwrt uses proprietary drivers and supports
       | some routers with broadcom chips.
        
         | buescher wrote:
         | Chips - most chips - almost never fail from the aging
         | mechanisms described in the article. The folks interviewed are
         | not kidding when they are talking about applications with
         | extended lifetimes past ten years and only seeing these issues
         | at 28nm and below. In your typical laptop or whatever something
         | else breaks first anyway. There is a lot else that can break in
         | any piece of consumer electronics.
        
       | buescher wrote:
       | It's wild to see electromigration and temperature-driven aging
       | come back as a concern at today's dense process nodes. Pecht's
       | landmark work in the early nineties mostly debunked the old
       | handbook models that were based on those causes of failure. Most
       | electronics will still fail due to overstress or mechanical
       | fatigue from vibration or temperature cycling, of course, but
       | it's really interesting to see these concerns return.
        
       ___________________________________________________________________
       (page generated 2021-03-15 23:02 UTC)